[gmx-users] Re: Simluations in vacuum - energy increase

[Erik Marklund]

David van der Spoel skrev 2011-05-01 20.10:
> On 2011-05-01 20.00, Zoe Hall wrote:
>> Hi - thanks - much improved with shorter timestep and increase in 
>> constraint accuracy. Manual says "hbonds", not "h-bonds", however 
>> didn't make much difference which one I used. Could someone suggest 
>> the reason why you would want to turn temperature coupling off during 
>> the simulations - I don't really understand this.
>>
> In our work we want to study evaporation and cooling induced by the 
> leaving water molecules. If you don't have water molecules, or even 
> assume some kind of heat bath (e.g. a gas atmosphere), you could turn 
> on T-coupling.
We furthermore managed to simulate lysozyme in vacuo with 1 fs time 
steps, without energy drift. See Marklund et. al. PCCP v. 11:36, pp. 
7741 (2009) for clues.
>> Thanks,
>>
>> Zoe
>>
>> On 2011-04-30 14.17, Zoe Hall wrote:
>>> Gmx-users,
>>>
>>> I am trying to carry out a simulation of lysozyme in vacuo using the
>>> OPLS-AA forcefield. After energy minimisation, the protein is run for
>>> 10ps with position restraints and temperature coupling on. This is
>>> followed by the full production run of 10ns with temperature and
>>> pressure coupling turned off, H bonds are constrained using LINCS and a
>>> time step of 1fs. For vacuum conditions, the periodic boundary
>>> conditions are turned off, and no cut-offs are used. When I carry out
>>> the 10ns simulation the total energy gradually increases, as does the
>>> temperature from 300 to 500K. I presume this is because the temperature
>>> coupling is turned off, however that is what I have noted from the
>>> literature that others do for their vacuum simulations. Can anyone shed
>>> any light on this? Following is my method.
>>>
>>> integrator = md
>>>
>>> tinit = 0
>>>
>>> dt = 0.001
>>>
>>> nsteps = 10000000
>>>
>>> nstxout = 20000
>>>
>>> nstvout = 20000
>>>
>>> nstfout = 0
>>>
>>> nstlog = 100000
>>>
>>> nstenergy = 100000
>>>
>>> nstxtcout = 20000
>>>
>>> energygrps = Protein
>>>
>>> nstcomm = 5
>>>
>>> nstlist = 0
>>>
>>> ns-type = simple
>>>
>>> pbc = no
>>>
>>> rlist = 0
>>>
>>> coulombtype = Cut-off
>>>
>>> rcoulomb = 0
>>>
>>> epsilon_r = 2
>>>
>>> vdw-type = Cut-off
>>>
>>> rvdw =0
>>>
>>> Tcoupl = no
>>>
>>> tc-grps = Protein
>>>
>>> tau_t = 0.1
>>>
>>> ref_t = 300
>>>
>>> Pcoupl = no
>>>
>>> Pcoupltype = Isotropic
>>>
>>> tau_p = 1
>>>
>>> compressibility = 4.5e-5
>>>
>>> ref_p = 1.0
>>>
>>> gen_vel = yes ;
>>>
>>> gen_temp = 300.0
>>>
>>> gen_seed = -1
>>>
>>> constraints = hbonds
>>>
>>> constraint-algorithm = LINCS
>>>
>>> lincs_order = 4
>>>
>>> Thanks,
>>>
>>> Zoe
>>>
>>> Zoe Hall
>>>
>>> Department of Chemistry
>>>
>>> Oxford University
>>>
>>> _zoe.hall at chem.ox.ac.uk_
>>>
>> Are you sure h-bonds are being constrained, because otherwise the time
>> step is too large? (maybe you need to write h-bonds). You may need to
>> increase the constraint accuracy as well. We did all our vacuum calcs in
>> double precision as well IIRC.
>>
>
>


-- 
-----------------------------------------------
Erik Marklund, PhD student
Dept. of Cell and Molecular Biology, Uppsala University.
Husargatan 3, Box 596,    75124 Uppsala, Sweden
phone:    +46 18 471 4537        fax: +46 18 511 755
erikm at xray.bmc.uu.se    http://folding.bmc.uu.se/